No etcd snapshots or backup strategy

[letan-assistant]

There is no backup mechanism for cluster state or persistent data:

• No etcd snapshots: k3s supports automatic snapshots via --etcd-snapshot-schedule-cron and off-site storage via --etcd-s3, but neither is configured.
• No block storage backup: The 10GB Vultr volume attached to the control plane (/var/lib/rancher/k3s) has no snapshot or backup schedule.
• Destructive IP-change behavior: k3s-init.sh wipes /var/lib/rancher/k3s/server/db when the VPC IP changes, with no prior backup.

A control plane failure or volume corruption currently means total, unrecoverable cluster state loss.

Suggested approach:

• Enable k3s automatic etcd snapshots with a cron schedule
• Configure S3-compatible off-site snapshot storage (Vultr Object Storage or similar)
• Consider Vultr block storage snapshots via API or Terraform
• Document a disaster recovery runbook